test(deviceauth): port device auth tests to integration:ng by p-targowicz · Pull Request #2074 · mendersoftware/mender-server

p-targowicz · 2026-07-17T10:06:10Z

Part of the suite-by-suite python-to-Go port (see #2079). Ports
test_devauth.py: preauthorization, device and authset management, auth
requests across RSA, four ECDSA curves and Ed25519. The python tests
keep running in parallel. Builds on #2079.

codecov · 2026-07-17T11:56:20Z

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

bahaa-ghazal

looks good, would appreciate if could use the same style as in here

bahaa-ghazal

LGTM!
I would also like @merlin-northern to take a look at it to be more safe I didn't miss anything

p-targowicz · 2026-07-29T09:22:59Z

LGTM! I would also like @merlin-northern to take a look at it to be more safe I didn't miss anything

absolutely @merlin-northern please take a look whenever you have some spare time 🙏

Preauthorization, device and authset management, and the auth request flow across all six key types, ported from test_devauth.py, which keeps running in parallel. Signed-off-by: Patryk Targowicz <patryk.targowicz@northern.tech>

merlin-northern

okey, I would like to discuss during the MSTM the policy on rewriting the tests. let's have a decision and a defined approach. one way is: leave the python tests as legacy, develop new tests in the new framework, and every time we modify something we write/update the test case in tests-ng. also I would like to discuss the approach to AI generated code and set the policy for that.

I am not blocking this, but I would kindly request no more porting before we discuss&agree as mentioned above.

merlin-northern · 2026-08-05T07:00:58Z

+	"github.com/mendersoftware/mender-server/tests/runner/tests/common"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"github.com/stretchr/testify/suite"


Suggested change

"github.com/stretchr/testify/suite"

"github.com/google/uuid"

"github.com/stretchr/testify/assert"

"github.com/stretchr/testify/require"

"github.com/stretchr/testify/suite"

"github.com/mendersoftware/mender-server/pkg/api/client"

"github.com/mendersoftware/mender-server/pkg/utils/types"

"github.com/mendersoftware/mender-server/services/deviceauth/model"

"github.com/mendersoftware/mender-server/tests/runner/tests/common"

nitpicking.

merlin-northern · 2026-08-05T08:12:14Z

+	s.Run("Ok", s.preauthOk)
+	s.Run("FailDuplicate", s.preauthFailDuplicate)
+	s.Run("FailBadRequest", s.preauthFailBadRequest)


looking for instance at the compareDevs and preauthOk it is impossible to distinguish between the test case and the helper function. there should be a clean distinction.

merlin-northern · 2026-08-05T08:20:56Z

+	KP     *common.KeyPair
+}
+
+type devFixture struct {


why is it called a fixture?

merlin-northern · 2026-08-05T14:27:31Z

+// docker_compose_environment.go), so callers that need to compare the full
+// listing/paging/count against a locally tracked device set must pass ids
+// to filter out devices left behind by other tests/suites.
+func (s *DevauthManagementV2Suite) listDevices(ctx context.Context, status string, page, perPage int32, ids ...string) ([]client.Device, error) {


we have been there. the python tests have 10 tons of helpers, abstraction, and stuff that mask the true purpose of the test and makes it impossible to read now and insanely difficult for future us to maintain. this is what the python tests proved: the insane design patterns and strange abstractions have no place in the test code. the test code should be an easy to read and super simple to understand reference for how the product is expected to work, even so at the cost of duplicating code. what we did with the tests-ng was to write the test cases calling directly the pkg/client everywhere where possible (AcceptWait func is one important exception, about which below). this makes the test maintainable, why? because in 10 years someone reading the test code sees DeviceAuthManagementListDevices().Execute() and immediately sees what is being tested, and why, without jumps to definitions in other places. there are cases where we need to have a common code as is the case with accepting a device (see AcceptWait) so some balance is eeded.

merlin-northern · 2026-08-05T14:28:29Z

+func (s *DevauthManagementV2Suite) getDevice(ctx context.Context, id string) (*client.Device, error) {
+	dev, _, err := s.APIClient.DeviceAuthenticationManagementAPIAPI.
+		DeviceAuthManagementGetDevice(ctx, id).Execute()
+	return dev, err
+}


see below, this is another helper that makes things worse.

merlin-northern · 2026-08-05T18:35:40Z

+	// object. The generated client can't build this (identity_data is
+	// typed as an object), so this one goes over a raw request.
+	body, err := json.Marshal(map[string]any{
+		"identity_data": `{"mac": "foo"}`,


this really looks like a trap and a riddle, takes some time to understand that the `` make all the difference.

merlin-northern · 2026-08-05T18:43:10Z

+		"pubkey":        kp.PublicKeyPEM(),
+	})
+	require.NoError(err)
+	resp, err := common.RawRequest(ctx, s.APIClient, http.MethodPost, "/api/management/v2/devauth/devices", body)


hardcoded endpoint, and method, has nothing to do with the code or spec. I think the correct way to inject the bad payload is to override RoundTrip in the http.RoundTripper interface and set a new HTTPClient in the openapi client configuration with custom Transport.

p-targowicz changed the base branch from rewrite-backend-integration-tests-ng to ng-rewrite-01-foundation July 17, 2026 10:20

p-targowicz force-pushed the ng-rewrite-01-foundation branch from d19a365 to 0754f85 Compare July 17, 2026 10:34

p-targowicz force-pushed the ng-rewrite-03-devauth branch from 4098ae5 to c6c45f3 Compare July 17, 2026 10:34

p-targowicz changed the base branch from ng-rewrite-01-foundation to ng-rewrite-02-health July 17, 2026 10:35

p-targowicz marked this pull request as ready for review July 17, 2026 11:09

p-targowicz force-pushed the ng-rewrite-02-health branch from 294aafd to 75cde9a Compare July 20, 2026 09:23

p-targowicz force-pushed the ng-rewrite-03-devauth branch from c6c45f3 to dd74371 Compare July 20, 2026 09:23

p-targowicz force-pushed the ng-rewrite-02-health branch from 75cde9a to 4214e83 Compare July 20, 2026 10:07

p-targowicz force-pushed the ng-rewrite-03-devauth branch from dd74371 to f3755c7 Compare July 20, 2026 10:09

bahaa-ghazal self-requested a review July 27, 2026 13:26

bahaa-ghazal requested changes Jul 27, 2026

View reviewed changes

p-targowicz force-pushed the ng-rewrite-03-devauth branch from f3755c7 to c62261f Compare July 28, 2026 07:13

Base automatically changed from ng-rewrite-02-health to ng-rewrite-01f-inventory-v2 July 28, 2026 09:07

p-targowicz force-pushed the ng-rewrite-01f-inventory-v2 branch from d905f66 to be3ac45 Compare July 28, 2026 09:19

p-targowicz force-pushed the ng-rewrite-03-devauth branch 4 times, most recently from 781170a to 90db24e Compare July 29, 2026 07:48

bahaa-ghazal reviewed Jul 29, 2026

View reviewed changes

bahaa-ghazal requested a review from merlin-northern July 29, 2026 08:52

p-targowicz force-pushed the ng-rewrite-01f-inventory-v2 branch from 125cadc to 6d91e79 Compare July 29, 2026 13:03

p-targowicz force-pushed the ng-rewrite-03-devauth branch from 90db24e to 35b08e7 Compare July 29, 2026 13:03

Base automatically changed from ng-rewrite-01f-inventory-v2 to main July 29, 2026 20:17

test(deviceauth): port device auth tests to integration:ng

d1c4932

Preauthorization, device and authset management, and the auth request flow across all six key types, ported from test_devauth.py, which keeps running in parallel. Signed-off-by: Patryk Targowicz <patryk.targowicz@northern.tech>

p-targowicz force-pushed the ng-rewrite-03-devauth branch from 35b08e7 to d1c4932 Compare August 3, 2026 10:14

bahaa-ghazal approved these changes Aug 3, 2026

View reviewed changes

merlin-northern reviewed Aug 5, 2026

View reviewed changes

Uh oh!

Conversation

p-targowicz commented Jul 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov Bot commented Jul 17, 2026

Codecov Report

Uh oh!

bahaa-ghazal left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

bahaa-ghazal left a comment

Choose a reason for hiding this comment

Uh oh!

p-targowicz commented Jul 29, 2026

Uh oh!

merlin-northern left a comment

Choose a reason for hiding this comment

Uh oh!

merlin-northern Aug 5, 2026

Choose a reason for hiding this comment

Uh oh!

merlin-northern Aug 5, 2026

Choose a reason for hiding this comment

Uh oh!

merlin-northern Aug 5, 2026

Choose a reason for hiding this comment

Uh oh!

merlin-northern Aug 5, 2026

Choose a reason for hiding this comment

Uh oh!

merlin-northern Aug 5, 2026

Choose a reason for hiding this comment

Uh oh!

merlin-northern Aug 5, 2026

Choose a reason for hiding this comment

Uh oh!

merlin-northern Aug 5, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

p-targowicz commented Jul 17, 2026 •

edited

Loading